This application is related to a number of applications filed on even date herewith for the same inventors, bearing the titles: "Planar Reflective Optical Devices", "Integration of Free-Space Planar Optical Components", and "Integrated Optics Facsimile Apparatus".
This relates to integrated optics for free space computing environments. More particularly, this relates to techniques for developing interacting arrays of optical elements that need not be aligned with cumbersome mechanical arrangements in order to realize free space computing.
Current optical systems employ a variety of optical components, such as lenses and beam splitters, which are mounted and aligned by mechanical means. In complex optical systems which consist of many individual components, the alignment and the stability of these components become critical issues. The alignment problem becomes particularly severe when the required precision approaches the limits of conventional fine mechanics.
In free-space optical computing systems the required precision is related to the size of the optical logic gates which are used to perform logic operations, such as AND and OR functions. Typically, the sizes of the optical logic devices that are currently investigated are in the range of a few microns or less. This means that the alignment tolerances for the optical components have to be in the submicron range. Conventional mechanical alignment of optical components is possible with a precision in the range of 10 microns. Below that range, conventional techniques become increasingly expensive. This means that for the purpose of building complex optical systems which require very precise mounting, alternative technique must be devised. The issue, then, is "How do we put all the components together with a submicron precision such that the resulting arrangement will stay stable despite influences such as temperature changes, mechanical stress, aging effects, etc.".
In the semiconductor art, interconnections also present a challenge. The challenge in the semiconductor art results from the fact that electronic elements (e.g. transistors, resistors, paths, etc.) are basically planar devices and the interconnections between the electronic elements are restricted to planar paths. This is particularly troublesome for connections to and from the integrated circuits. The reason for that relates to the need to reach an edge of the integrated circuit and the need to provide sufficient current carrying capabilities to drive the signals on the input/output leads to the desired voltage levels in the presence of circuit capacitances.
To solve the interconnection problem in electronic circuits recent proposals suggest the use of free-space optical means. The configurations for optical interconnections of VLSI systems that are most relevant to this invention were described by Goodman et al. in Proceeding of the IEEE Vol. 72, No. 7, Jul. 1984, pp. 850-866, and more recently, by Brenner et al. in Applied Optics Vol. 27, No. 20, 15 Oct., 1988, pp. 4251-4254. Both publications describe arrangements where the exposed surface of an integrated circuit contains optical detectors. In the Goodman et al. arrangements light is applied either from outside the integrated circuit or from light sources at the edges of the integrated circuit. The light is directed to a hologram substrate situated above the integrated circuit's surface, and the hologram routes the optical signals to the desired detectors, based on the information imbedded in the hologram.
The Brenner et al. arrangement is similar to that of Goodman et al. in that both the light source and the detector are on the same exposed integrated circuit surface. The light is directed upwards toward a substrate that is held above the integrated circuit's exposed surface and which contains holograms at designated locations of the substrate. The holograms are at the surface of the substrate that faces the integrated circuit (the near-end surface). A mirrored surface is placed at the opposite surface of the substrate (the far-end surface). The light from an integrated circuit light source passes through the hologram of the substrate, reflects off the mirrored surface, and returns to the surface of the integrated circuit where an optical detector is situated (if the arrangement is aligned properly). In one implementation of the Brenner et al. arrangement, the light that is reflected off the far-end surface is reflected off a mirrored portion on the near-end surface, and is reflected again off the far-end surface before it reaches the integrated circuit. This allows a lengthening of the optical path and offers certain flexibility in the positioning of the light detector vis-a-vis the light source.
The above arrangements perhaps solve the problem of relaying signals to and from integrated circuits. They do not, however, solve the needs of an optical computing system. First, they do not obviate the need for alignment. The holograms and the mirrors in the above-described arrangements must be aligned precisely. Second, they address only the relatively simple process of sending an optical signal from one point and receiving it at another. They do not solve the problem of creating a circuit having optical elements that more generally interact with one another. Third, the above-described systems basically deal with one-to-one or one-to-many communications, whereas optical computing applications need to relay images (a collection of spots). The importance of this difference lies in the fact that a lens inverts the image. In the case of spot to spot communication that inversion is irrelevant to the detector and thus not considered. In the case of image manipulations, in contradistinction, one may not invert portions of the image with impunity.
Directing attention to the semiconductor integrated circuit manufacturing art, it may be observed that current techniques create entire circuits on a single substrate, with the resulting attribute that the interconnection fabric for the manufactured elements is created concurrently with the circuit elements themselves. Also, the relative positioning of the circuit elements, relative to one another, is also fixed.
The latter is not necessarily important in electronic circuit embodiments because the interconnections are physical, point-to-point interconnections, within the circuit's structure. It could be vitally important in optical circuits. In the context of this invention, an optical circuit is any arrangement of optical elements such as mirrors, lenses, etc. that perform predefined transformations on optical signals.
Artisans have realized the benefits of batch semiconductor fabrication techniques for optical circuits. For example, in "Fabrication of Planar Optical Phase Elements," Optical Communications, Vol. 8 No. 2 Jun. 1973, pp. 160-162, Firester et al. described a technique for fabricating optical phase elements using relatively conventional fabrication techniques. Their process starts with glass covered with evaporated aluminum. The aluminum is coated with a photoresist, a binary pattern mask is applied to the photoresist, and the aluminum is chemically etched to the pattern delineated by the mask. Thereafter, the remaining photoresist is removed and a layer of thorium fluoride is deposited on top using resistance heated vacuum evaporation. Finally, the aluminum pattern is chemically etched away, leaving the thorium fluoride that was in contact with the glass.
More recently, in "Rectangular-apertured micro-Fresnel lens array fabricated by electron-beam lithography" Applied Optics, Vol. 26, No. 3, Feb. 1987, pp. 587-591, Shiono at al. described an electron beam approach for creating an array of Fresnel lenses where the effective etching of a coating baked onto a glass substrate is accomplished with an electronbeam writing system.
These articles demonstrate the use of batch processing techniques to create a plurality of optical elements that in combination form an array of elements. Such arrays are used to effect a particular optical element, such as a lens array, a detector array, a hologram, etc. A common characteristic of these array-elements is that energy is extracted from, or applied to, the entire manufactured array of devices as a unit. There is no uniquely designed interaction between the elements themselves that would convert their particular arrangement into a "circuit" rather than a "macro" optical element.
Thus, an unsolved need still exists for creating optical computing circuits that do not require stringent mechanical alignment.